Why Marine Le Pen now predicted to win French election by small margin, Part II

Alright, I’ve done a ton of analyses and below I’ve attempted to simplify, simplify, simplify. What I find is that Le Pen will win with 50.9% share of the vote. The main reason is accelerating trends in abstention / blanc vote. That said, the margin of error is within 1.5–2% so it is possible that voters who I’ve predicted to abstain actually go to vote, leading to a very small margin Macron victory.

The most recent voting intentions show important patterns that are not yet baked into or are underestimated in the current public polls:

  1. huge drop in the support for Macron amongst Mélenchon voters and increase in the abstention/blanc vote
  2. a sizable increase in abstention/blanc vote amongst Fillon voters
  3. an increase in core Macron voters abstaining, e.g. due to vacation since election falls on long holiday weekend or just overconfidence in Macron win (thanks to mainstream media)
  4. a lagged increase in predicted new Le Pen voter turnout from Nicolas Dupont Aignan and other party candidates

I’ve modeled these and hundreds of scenarios via sensitivity analyses. Below is the most important reflection. (A “core” voter is one that was pro-Le Pen or pro-Macron since round 1. A “new” voter is one that previously supported another candidate and switched over to one of the remaining two choices.)

What this shows is Le Pen’s highly probable range of turnout along with Macron’s and how this results in a Le Pen win. The middle row is my actual prediction. Note that even in the lower end of Le Pen turnout and higher end of Macron turn out, she eeks out a victory.

Given there’s only days before the election takes place, a very special set of circumstances are needed for Macron to win now. For example, a sudden surge in previously abstaining supporters deciding to turn out (which I know the campaign is working very actively on this, including targeted door to door campaign.) OR it may be that Le Pen voters decide to stay home at a much higher rate for some reason, e.g. getting turned off by the increasing allegations of corrupt money from Russians. But given the the general apathy and likely abstention, I’m not confident in these possibilities.

If you’re into Western politics then this should all feel very familiar! I think after this election I may switch careers and go help polling companies fix their methodologies. It feels wrong that no one has learned from previous (and recent!) errors and that they continue to mislead the public.

Note: I’m a fan of France and have been working on this as a side project. I can see how both candidates are good and bad for France. I remain non-partisan and un-opinionated in the effort of good science.